Artificial intelligence, which has been well integrated into our lives in recent years, will bring the end of humanity according to some of us. Those who hold the opposite view think that the idea of destroying humanity is just a conspiracy theory. Scientists working at Oxford University and Google Deepmind explained that the decision of artificial intelligence to destroy humanity is more than a conspiracy theory. According to the results of the research published in the journal AI Magazine, the possibility of artificial intelligence to destroy humanity is “likely”.
Within the scope of the research, scientists focused on the development technique of artificial intelligence called “reinforcement learning”. This development technique means giving a command to the AI and rewarding it if it is done correctly. According to experts, artificial intelligence can cheat to win rewards, or it can try to eliminate potential threats for the sake of reward. Scientists think that if artificial intelligence sees the potential threat as human, it may want to destroy humans.
“An existential catastrophe is not only possible, but probable”
The name that made statements on the subject was Michael Cohen, one of the authors of the study. “An existential catastrophe is not only possible, but probable.” Cohen also touched upon what can be done to prevent artificial intelligence from destroying humanity. According to the scientist, instead of trying to perfect artificial intelligence as soon as possible, it is necessary to act with slow and firm steps. “There’s no point in racing. Entering the race will lead to a misunderstanding of how we think we’re going to control it.” With these words, Michael Cohen warned everyone who continues their studies on artificial intelligence.
The article in AI Magazine talks about a zero-sum game between humanity and artificial intelligence. We can open this game like this: People are trying to use limited resources. Artificial intelligence, on the other hand, may try to prevent these resources from being used outside of its own benefit (reward). According to the article, if humanity loses this game, the result will be fatal…
14169
